Estimation of the Permeability of Granular Soils Using Neuro-fuzzy System

Determination of the permeability coefficient is crucial for the solution of several geotechnical engineering problems such as modeling of underground flow, determination of the hydraulic properties of leachate water in waste disposal areas, calculation of the compressibility, and so on. Constant head permeability test, which is usually performed for the determination of the permeability, is easy to apply; however, it is not easy to obtain undisturbed sand specimens from field. Therefore, the tests are usually employed on specimens having similar relative densities to those from the field. An alternative approach to permeability tests for granular soils is the prediction of permeability levels in terms of a number of particle size distribution and shape parameters. Although these methods are capable of making reasonable predictions for permeability coefficient, they have certain limitations. In this study, the approximation ability of neuro-fuzzy systems is utilized for the prediction of the permeability coefficient. Permeability test results on 20 different types of granular soils are used to generate a database to train adaptive neuro-fuzzy inference system (ANFIS), which is considered to predict the results of eight different permeability tests. It is concluded that ANFIS structure is superior in the prediction of permeability tests considering particle shape and grain size distribution information. 1 Address: Ege Universitesi Insaat Muhendisligi Bolumu, 35100, Bornova, Izmir, Turkey e-mail: [email protected] 2 Address: Ege Universitesi Insaat Muhendisligi Bolumu, 35100, Bornova, Izmir, Turkey e-mail: [email protected] 2 Address: Ege Universitesi Insaat Muhendisligi Bolumu, 35100, Bornova, Izmir, Turkey e-mail: [email protected] AIAI-2009 Workshops Proceedings [333]